Latest Report | Algorithm Submission
[2021-07-13] We have released a new report from the Face Recognition Vendor Test (FRVT) program: Part 7: Identification for Paperless Travel and Immigration. This report will be updated on a regular basis as new analyses are implemented, and as results for newly submitted FRVT algorithms are produced. The press release can be found here.
One-to-many facial recognition has potential in the airport transit setting where a traveler’s face can be matched against galleries of individuals expected to be present. This is done in the context of paperless travel where the traditional boarding pass is replaced with the presentation of biometrics to a camera. The primary considerations for this case is as a double-duty for access control to an aircraft and at facilitation of recording a visa-holder’s departure from a country. This experiment aims to quantify face matching errors by simulating departing flights and centralized airport checkpoints.
[2021-07-13] The table below shows the False Negative Identification “miss” Rates (FNIR), i.e. the percentage of travelers not matched to their gallery photo(s). This is done where a threshold is set to limit the False Positive Identification Rate (FPIR) to 0.003. In this leaderboard, the 420 person galleries represent aircraft boarding while the 42000 represents an airport security line where many more people are expected. The k value represents the number of images of each enrollee in each gallery.
Subscribe to the FRVT mailing list to receive emails when announcements or updates are made.